Merge branch 'js/shortlog'
[git] / builtin-grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #include "cache.h"
7 #include "blob.h"
8 #include "tree.h"
9 #include "commit.h"
10 #include "tag.h"
11 #include "tree-walk.h"
12 #include "builtin.h"
13 #include <regex.h>
14 #include "grep.h"
15 #include <fnmatch.h>
16 #include <sys/wait.h>
17
18 /*
19  * git grep pathspecs are somewhat different from diff-tree pathspecs;
20  * pathname wildcards are allowed.
21  */
22 static int pathspec_matches(const char **paths, const char *name)
23 {
24         int namelen, i;
25         if (!paths || !*paths)
26                 return 1;
27         namelen = strlen(name);
28         for (i = 0; paths[i]; i++) {
29                 const char *match = paths[i];
30                 int matchlen = strlen(match);
31                 const char *cp, *meta;
32
33                 if (!matchlen ||
34                     ((matchlen <= namelen) &&
35                      !strncmp(name, match, matchlen) &&
36                      (match[matchlen-1] == '/' ||
37                       name[matchlen] == '\0' || name[matchlen] == '/')))
38                         return 1;
39                 if (!fnmatch(match, name, 0))
40                         return 1;
41                 if (name[namelen-1] != '/')
42                         continue;
43
44                 /* We are being asked if the directory ("name") is worth
45                  * descending into.
46                  *
47                  * Find the longest leading directory name that does
48                  * not have metacharacter in the pathspec; the name
49                  * we are looking at must overlap with that directory.
50                  */
51                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
52                         char ch = *cp;
53                         if (ch == '*' || ch == '[' || ch == '?') {
54                                 meta = cp;
55                                 break;
56                         }
57                 }
58                 if (!meta)
59                         meta = cp; /* fully literal */
60
61                 if (namelen <= meta - match) {
62                         /* Looking at "Documentation/" and
63                          * the pattern says "Documentation/howto/", or
64                          * "Documentation/diff*.txt".  The name we
65                          * have should match prefix.
66                          */
67                         if (!memcmp(match, name, namelen))
68                                 return 1;
69                         continue;
70                 }
71
72                 if (meta - match < namelen) {
73                         /* Looking at "Documentation/howto/" and
74                          * the pattern says "Documentation/h*";
75                          * match up to "Do.../h"; this avoids descending
76                          * into "Documentation/technical/".
77                          */
78                         if (!memcmp(match, name, meta - match))
79                                 return 1;
80                         continue;
81                 }
82         }
83         return 0;
84 }
85
86 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
87 {
88         unsigned long size;
89         char *data;
90         char type[20];
91         char *to_free = NULL;
92         int hit;
93
94         data = read_sha1_file(sha1, type, &size);
95         if (!data) {
96                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
97                 return 0;
98         }
99         if (opt->relative && opt->prefix_length) {
100                 static char name_buf[PATH_MAX];
101                 char *cp;
102                 int name_len = strlen(name) - opt->prefix_length + 1;
103
104                 if (!tree_name_len)
105                         name += opt->prefix_length;
106                 else {
107                         if (ARRAY_SIZE(name_buf) <= name_len)
108                                 cp = to_free = xmalloc(name_len);
109                         else
110                                 cp = name_buf;
111                         memcpy(cp, name, tree_name_len);
112                         strcpy(cp + tree_name_len,
113                                name + tree_name_len + opt->prefix_length);
114                         name = cp;
115                 }
116         }
117         hit = grep_buffer(opt, name, data, size);
118         free(data);
119         free(to_free);
120         return hit;
121 }
122
123 static int grep_file(struct grep_opt *opt, const char *filename)
124 {
125         struct stat st;
126         int i;
127         char *data;
128         if (lstat(filename, &st) < 0) {
129         err_ret:
130                 if (errno != ENOENT)
131                         error("'%s': %s", filename, strerror(errno));
132                 return 0;
133         }
134         if (!st.st_size)
135                 return 0; /* empty file -- no grep hit */
136         if (!S_ISREG(st.st_mode))
137                 return 0;
138         i = open(filename, O_RDONLY);
139         if (i < 0)
140                 goto err_ret;
141         data = xmalloc(st.st_size + 1);
142         if (st.st_size != xread(i, data, st.st_size)) {
143                 error("'%s': short read %s", filename, strerror(errno));
144                 close(i);
145                 free(data);
146                 return 0;
147         }
148         close(i);
149         if (opt->relative && opt->prefix_length)
150                 filename += opt->prefix_length;
151         i = grep_buffer(opt, filename, data, st.st_size);
152         free(data);
153         return i;
154 }
155
156 static int exec_grep(int argc, const char **argv)
157 {
158         pid_t pid;
159         int status;
160
161         argv[argc] = NULL;
162         pid = fork();
163         if (pid < 0)
164                 return pid;
165         if (!pid) {
166                 execvp("grep", (char **) argv);
167                 exit(255);
168         }
169         while (waitpid(pid, &status, 0) < 0) {
170                 if (errno == EINTR)
171                         continue;
172                 return -1;
173         }
174         if (WIFEXITED(status)) {
175                 if (!WEXITSTATUS(status))
176                         return 1;
177                 return 0;
178         }
179         return -1;
180 }
181
182 #define MAXARGS 1000
183 #define ARGBUF 4096
184 #define push_arg(a) do { \
185         if (nr < MAXARGS) argv[nr++] = (a); \
186         else die("maximum number of args exceeded"); \
187         } while (0)
188
189 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
190 {
191         int i, nr, argc, hit, len, status;
192         const char *argv[MAXARGS+1];
193         char randarg[ARGBUF];
194         char *argptr = randarg;
195         struct grep_pat *p;
196
197         if (opt->extended || (opt->relative && opt->prefix_length))
198                 return -1;
199         len = nr = 0;
200         push_arg("grep");
201         if (opt->fixed)
202                 push_arg("-F");
203         if (opt->linenum)
204                 push_arg("-n");
205         if (!opt->pathname)
206                 push_arg("-h");
207         if (opt->regflags & REG_EXTENDED)
208                 push_arg("-E");
209         if (opt->regflags & REG_ICASE)
210                 push_arg("-i");
211         if (opt->word_regexp)
212                 push_arg("-w");
213         if (opt->name_only)
214                 push_arg("-l");
215         if (opt->unmatch_name_only)
216                 push_arg("-L");
217         if (opt->count)
218                 push_arg("-c");
219         if (opt->post_context || opt->pre_context) {
220                 if (opt->post_context != opt->pre_context) {
221                         if (opt->pre_context) {
222                                 push_arg("-B");
223                                 len += snprintf(argptr, sizeof(randarg)-len,
224                                                 "%u", opt->pre_context);
225                                 if (sizeof(randarg) <= len)
226                                         die("maximum length of args exceeded");
227                                 push_arg(argptr);
228                                 argptr += len;
229                         }
230                         if (opt->post_context) {
231                                 push_arg("-A");
232                                 len += snprintf(argptr, sizeof(randarg)-len,
233                                                 "%u", opt->post_context);
234                                 if (sizeof(randarg) <= len)
235                                         die("maximum length of args exceeded");
236                                 push_arg(argptr);
237                                 argptr += len;
238                         }
239                 }
240                 else {
241                         push_arg("-C");
242                         len += snprintf(argptr, sizeof(randarg)-len,
243                                         "%u", opt->post_context);
244                         if (sizeof(randarg) <= len)
245                                 die("maximum length of args exceeded");
246                         push_arg(argptr);
247                         argptr += len;
248                 }
249         }
250         for (p = opt->pattern_list; p; p = p->next) {
251                 push_arg("-e");
252                 push_arg(p->pattern);
253         }
254
255         /*
256          * To make sure we get the header printed out when we want it,
257          * add /dev/null to the paths to grep.  This is unnecessary
258          * (and wrong) with "-l" or "-L", which always print out the
259          * name anyway.
260          *
261          * GNU grep has "-H", but this is portable.
262          */
263         if (!opt->name_only && !opt->unmatch_name_only)
264                 push_arg("/dev/null");
265
266         hit = 0;
267         argc = nr;
268         for (i = 0; i < active_nr; i++) {
269                 struct cache_entry *ce = active_cache[i];
270                 char *name;
271                 if (!S_ISREG(ntohl(ce->ce_mode)))
272                         continue;
273                 if (!pathspec_matches(paths, ce->name))
274                         continue;
275                 name = ce->name;
276                 if (name[0] == '-') {
277                         int len = ce_namelen(ce);
278                         name = xmalloc(len + 3);
279                         memcpy(name, "./", 2);
280                         memcpy(name + 2, ce->name, len + 1);
281                 }
282                 argv[argc++] = name;
283                 if (argc < MAXARGS && !ce_stage(ce))
284                         continue;
285                 status = exec_grep(argc, argv);
286                 if (0 < status)
287                         hit = 1;
288                 argc = nr;
289                 if (ce_stage(ce)) {
290                         do {
291                                 i++;
292                         } while (i < active_nr &&
293                                  !strcmp(ce->name, active_cache[i]->name));
294                         i--; /* compensate for loop control */
295                 }
296         }
297         if (argc > nr) {
298                 status = exec_grep(argc, argv);
299                 if (0 < status)
300                         hit = 1;
301         }
302         return hit;
303 }
304
305 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
306 {
307         int hit = 0;
308         int nr;
309         read_cache();
310
311 #ifdef __unix__
312         /*
313          * Use the external "grep" command for the case where
314          * we grep through the checked-out files. It tends to
315          * be a lot more optimized
316          */
317         if (!cached) {
318                 hit = external_grep(opt, paths, cached);
319                 if (hit >= 0)
320                         return hit;
321         }
322 #endif
323
324         for (nr = 0; nr < active_nr; nr++) {
325                 struct cache_entry *ce = active_cache[nr];
326                 if (!S_ISREG(ntohl(ce->ce_mode)))
327                         continue;
328                 if (!pathspec_matches(paths, ce->name))
329                         continue;
330                 if (cached) {
331                         if (ce_stage(ce))
332                                 continue;
333                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
334                 }
335                 else
336                         hit |= grep_file(opt, ce->name);
337                 if (ce_stage(ce)) {
338                         do {
339                                 nr++;
340                         } while (nr < active_nr &&
341                                  !strcmp(ce->name, active_cache[nr]->name));
342                         nr--; /* compensate for loop control */
343                 }
344         }
345         free_grep_patterns(opt);
346         return hit;
347 }
348
349 static int grep_tree(struct grep_opt *opt, const char **paths,
350                      struct tree_desc *tree,
351                      const char *tree_name, const char *base)
352 {
353         int len;
354         int hit = 0;
355         struct name_entry entry;
356         char *down;
357         int tn_len = strlen(tree_name);
358         char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
359
360         if (tn_len) {
361                 tn_len = sprintf(path_buf, "%s:", tree_name);
362                 down = path_buf + tn_len;
363                 strcat(down, base);
364         }
365         else {
366                 down = path_buf;
367                 strcpy(down, base);
368         }
369         len = strlen(path_buf);
370
371         while (tree_entry(tree, &entry)) {
372                 strcpy(path_buf + len, entry.path);
373
374                 if (S_ISDIR(entry.mode))
375                         /* Match "abc/" against pathspec to
376                          * decide if we want to descend into "abc"
377                          * directory.
378                          */
379                         strcpy(path_buf + len + entry.pathlen, "/");
380
381                 if (!pathspec_matches(paths, down))
382                         ;
383                 else if (S_ISREG(entry.mode))
384                         hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
385                 else if (S_ISDIR(entry.mode)) {
386                         char type[20];
387                         struct tree_desc sub;
388                         void *data;
389                         data = read_sha1_file(entry.sha1, type, &sub.size);
390                         if (!data)
391                                 die("unable to read tree (%s)",
392                                     sha1_to_hex(entry.sha1));
393                         sub.buf = data;
394                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
395                         free(data);
396                 }
397         }
398         return hit;
399 }
400
401 static int grep_object(struct grep_opt *opt, const char **paths,
402                        struct object *obj, const char *name)
403 {
404         if (obj->type == OBJ_BLOB)
405                 return grep_sha1(opt, obj->sha1, name, 0);
406         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
407                 struct tree_desc tree;
408                 void *data;
409                 int hit;
410                 data = read_object_with_reference(obj->sha1, tree_type,
411                                                   &tree.size, NULL);
412                 if (!data)
413                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
414                 tree.buf = data;
415                 hit = grep_tree(opt, paths, &tree, name, "");
416                 free(data);
417                 return hit;
418         }
419         die("unable to grep from object of type %s", typename(obj->type));
420 }
421
422 static const char builtin_grep_usage[] =
423 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
424
425 static const char emsg_invalid_context_len[] =
426 "%s: invalid context length argument";
427 static const char emsg_missing_context_len[] =
428 "missing context length argument";
429 static const char emsg_missing_argument[] =
430 "option requires an argument -%s";
431
432 int cmd_grep(int argc, const char **argv, const char *prefix)
433 {
434         int hit = 0;
435         int cached = 0;
436         int seen_dashdash = 0;
437         struct grep_opt opt;
438         struct object_array list = { 0, 0, NULL };
439         const char **paths = NULL;
440         int i;
441
442         memset(&opt, 0, sizeof(opt));
443         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
444         opt.relative = 1;
445         opt.pathname = 1;
446         opt.pattern_tail = &opt.pattern_list;
447         opt.regflags = REG_NEWLINE;
448
449         /*
450          * If there is no -- then the paths must exist in the working
451          * tree.  If there is no explicit pattern specified with -e or
452          * -f, we take the first unrecognized non option to be the
453          * pattern, but then what follows it must be zero or more
454          * valid refs up to the -- (if exists), and then existing
455          * paths.  If there is an explicit pattern, then the first
456          * unrecognized non option is the beginning of the refs list
457          * that continues up to the -- (if exists), and then paths.
458          */
459
460         while (1 < argc) {
461                 const char *arg = argv[1];
462                 argc--; argv++;
463                 if (!strcmp("--cached", arg)) {
464                         cached = 1;
465                         continue;
466                 }
467                 if (!strcmp("-a", arg) ||
468                     !strcmp("--text", arg)) {
469                         opt.binary = GREP_BINARY_TEXT;
470                         continue;
471                 }
472                 if (!strcmp("-i", arg) ||
473                     !strcmp("--ignore-case", arg)) {
474                         opt.regflags |= REG_ICASE;
475                         continue;
476                 }
477                 if (!strcmp("-I", arg)) {
478                         opt.binary = GREP_BINARY_NOMATCH;
479                         continue;
480                 }
481                 if (!strcmp("-v", arg) ||
482                     !strcmp("--invert-match", arg)) {
483                         opt.invert = 1;
484                         continue;
485                 }
486                 if (!strcmp("-E", arg) ||
487                     !strcmp("--extended-regexp", arg)) {
488                         opt.regflags |= REG_EXTENDED;
489                         continue;
490                 }
491                 if (!strcmp("-F", arg) ||
492                     !strcmp("--fixed-strings", arg)) {
493                         opt.fixed = 1;
494                         continue;
495                 }
496                 if (!strcmp("-G", arg) ||
497                     !strcmp("--basic-regexp", arg)) {
498                         opt.regflags &= ~REG_EXTENDED;
499                         continue;
500                 }
501                 if (!strcmp("-n", arg)) {
502                         opt.linenum = 1;
503                         continue;
504                 }
505                 if (!strcmp("-h", arg)) {
506                         opt.pathname = 0;
507                         continue;
508                 }
509                 if (!strcmp("-H", arg)) {
510                         opt.pathname = 1;
511                         continue;
512                 }
513                 if (!strcmp("-l", arg) ||
514                     !strcmp("--files-with-matches", arg)) {
515                         opt.name_only = 1;
516                         continue;
517                 }
518                 if (!strcmp("-L", arg) ||
519                     !strcmp("--files-without-match", arg)) {
520                         opt.unmatch_name_only = 1;
521                         continue;
522                 }
523                 if (!strcmp("-c", arg) ||
524                     !strcmp("--count", arg)) {
525                         opt.count = 1;
526                         continue;
527                 }
528                 if (!strcmp("-w", arg) ||
529                     !strcmp("--word-regexp", arg)) {
530                         opt.word_regexp = 1;
531                         continue;
532                 }
533                 if (!strncmp("-A", arg, 2) ||
534                     !strncmp("-B", arg, 2) ||
535                     !strncmp("-C", arg, 2) ||
536                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
537                         unsigned num;
538                         const char *scan;
539                         switch (arg[1]) {
540                         case 'A': case 'B': case 'C':
541                                 if (!arg[2]) {
542                                         if (argc <= 1)
543                                                 die(emsg_missing_context_len);
544                                         scan = *++argv;
545                                         argc--;
546                                 }
547                                 else
548                                         scan = arg + 2;
549                                 break;
550                         default:
551                                 scan = arg + 1;
552                                 break;
553                         }
554                         if (sscanf(scan, "%u", &num) != 1)
555                                 die(emsg_invalid_context_len, scan);
556                         switch (arg[1]) {
557                         case 'A':
558                                 opt.post_context = num;
559                                 break;
560                         default:
561                         case 'C':
562                                 opt.post_context = num;
563                         case 'B':
564                                 opt.pre_context = num;
565                                 break;
566                         }
567                         continue;
568                 }
569                 if (!strcmp("-f", arg)) {
570                         FILE *patterns;
571                         int lno = 0;
572                         char buf[1024];
573                         if (argc <= 1)
574                                 die(emsg_missing_argument, arg);
575                         patterns = fopen(argv[1], "r");
576                         if (!patterns)
577                                 die("'%s': %s", argv[1], strerror(errno));
578                         while (fgets(buf, sizeof(buf), patterns)) {
579                                 int len = strlen(buf);
580                                 if (buf[len-1] == '\n')
581                                         buf[len-1] = 0;
582                                 /* ignore empty line like grep does */
583                                 if (!buf[0])
584                                         continue;
585                                 append_grep_pattern(&opt, xstrdup(buf),
586                                                     argv[1], ++lno,
587                                                     GREP_PATTERN);
588                         }
589                         fclose(patterns);
590                         argv++;
591                         argc--;
592                         continue;
593                 }
594                 if (!strcmp("--not", arg)) {
595                         append_grep_pattern(&opt, arg, "command line", 0,
596                                             GREP_NOT);
597                         continue;
598                 }
599                 if (!strcmp("--and", arg)) {
600                         append_grep_pattern(&opt, arg, "command line", 0,
601                                             GREP_AND);
602                         continue;
603                 }
604                 if (!strcmp("--or", arg))
605                         continue; /* no-op */
606                 if (!strcmp("(", arg)) {
607                         append_grep_pattern(&opt, arg, "command line", 0,
608                                             GREP_OPEN_PAREN);
609                         continue;
610                 }
611                 if (!strcmp(")", arg)) {
612                         append_grep_pattern(&opt, arg, "command line", 0,
613                                             GREP_CLOSE_PAREN);
614                         continue;
615                 }
616                 if (!strcmp("--all-match", arg)) {
617                         opt.all_match = 1;
618                         continue;
619                 }
620                 if (!strcmp("-e", arg)) {
621                         if (1 < argc) {
622                                 append_grep_pattern(&opt, argv[1],
623                                                     "-e option", 0,
624                                                     GREP_PATTERN);
625                                 argv++;
626                                 argc--;
627                                 continue;
628                         }
629                         die(emsg_missing_argument, arg);
630                 }
631                 if (!strcmp("--full-name", arg)) {
632                         opt.relative = 0;
633                         continue;
634                 }
635                 if (!strcmp("--", arg)) {
636                         /* later processing wants to have this at argv[1] */
637                         argv--;
638                         argc++;
639                         break;
640                 }
641                 if (*arg == '-')
642                         usage(builtin_grep_usage);
643
644                 /* First unrecognized non-option token */
645                 if (!opt.pattern_list) {
646                         append_grep_pattern(&opt, arg, "command line", 0,
647                                             GREP_PATTERN);
648                         break;
649                 }
650                 else {
651                         /* We are looking at the first path or rev;
652                          * it is found at argv[1] after leaving the
653                          * loop.
654                          */
655                         argc++; argv--;
656                         break;
657                 }
658         }
659
660         if (!opt.pattern_list)
661                 die("no pattern given.");
662         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
663                 die("cannot mix --fixed-strings and regexp");
664         compile_grep_patterns(&opt);
665
666         /* Check revs and then paths */
667         for (i = 1; i < argc; i++) {
668                 const char *arg = argv[i];
669                 unsigned char sha1[20];
670                 /* Is it a rev? */
671                 if (!get_sha1(arg, sha1)) {
672                         struct object *object = parse_object(sha1);
673                         if (!object)
674                                 die("bad object %s", arg);
675                         add_object_array(object, arg, &list);
676                         continue;
677                 }
678                 if (!strcmp(arg, "--")) {
679                         i++;
680                         seen_dashdash = 1;
681                 }
682                 break;
683         }
684
685         /* The rest are paths */
686         if (!seen_dashdash) {
687                 int j;
688                 for (j = i; j < argc; j++)
689                         verify_filename(prefix, argv[j]);
690         }
691
692         if (i < argc) {
693                 paths = get_pathspec(prefix, argv + i);
694                 if (opt.prefix_length && opt.relative) {
695                         /* Make sure we do not get outside of paths */
696                         for (i = 0; paths[i]; i++)
697                                 if (strncmp(prefix, paths[i], opt.prefix_length))
698                                         die("git-grep: cannot generate relative filenames containing '..'");
699                 }
700         }
701         else if (prefix) {
702                 paths = xcalloc(2, sizeof(const char *));
703                 paths[0] = prefix;
704                 paths[1] = NULL;
705         }
706
707         if (!list.nr)
708                 return !grep_cache(&opt, paths, cached);
709
710         if (cached)
711                 die("both --cached and trees are given.");
712
713         for (i = 0; i < list.nr; i++) {
714                 struct object *real_obj;
715                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
716                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
717                         hit = 1;
718         }
719         free_grep_patterns(&opt);
720         return !hit;
721 }